Related Experiment Video
Updated: Dec 17, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.2K
Automatic T Staging Using Weakly Supervised Deep Learning for Nasopharyngeal Carcinoma on MR Images
Qing Yang1,2, Ying Guo1,2, Xiaomin Ou1,2
1Department of Radiation Oncology, Fudan University Shanghai Cancer Center, Shanghai, China.
Journal of Magnetic Resonance Imaging : JMRI
|June 26, 2020
Summary
This study introduces a weakly-supervised deep learning method for automated nasopharyngeal carcinoma (NPC) T staging. The model achieved 75.59% accuracy and demonstrated good prognostic performance, aiding in cancer staging without extensive data annotation.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Deep learning shows promise for automated tumor staging.
- Automatic nasopharyngeal carcinoma (NPC) staging is challenging due to limited annotated datasets.
Purpose of the Study:
- To develop a weakly-supervised deep learning method for NPC T staging.
- To predict NPC T stage without requiring additional slice-level annotations.
Main Methods:
- A weakly-supervised deep learning network was developed using 1138 NPC cases.
- T1-weighted, T2-weighted, and contrast-enhanced T1-weighted images were utilized.
- Model performance was evaluated using receiver operating characteristic (ROC) curves and survival analysis (PFS, OS).
Main Results:
- The automated T staging model achieved 75.59% accuracy in the validation set.
- The average area under the ROC curve (AUC) was 0.943.
- No significant differences were observed in progression-free survival (PFS) and overall survival (OS) C-indexes compared to TNM staging (P > 0.05).
Conclusions:
- The weakly-supervised deep learning approach enables fully automated NPC T staging.
- The method demonstrates effective prognostic performance, comparable to traditional staging.
- This technique offers a viable solution for NPC staging with limited annotated data.
